Data from a recently conducted longitudinal caries study were used to assess the predictive value of specific groups of variables in predicting the level of dental caries incidence in adolescent subjects. The data available included demographic information, use of topical and systemic fluoride, initial caries scores, microbiologic levels of S. mutans and Lactobacilli on three occasions, and multiple 24-hour recall dietary intake information. These data were analyzed from two perspectives: 1) to assess the predictive value of predicting caries incidence over the whole range of caries experiences on the log scale; and 2) evaluate the predictive value of selected variables to identify subjects at high risk. The initial caries scores together with the microbiologic counts were the best predictors of caries incidence for the group. The initial caries scores, the microbiologic counts and the use of fluoride prophylaxes were the best fitting variables in the logistic models.